Discovery of a Novel IL-15 Based Protein with Improved
Total Page:16
File Type:pdf, Size:1020Kb
www.nature.com/scientificreports OPEN Discovery of a novel IL-15 based protein with improved developability and efcacy for Received: 1 November 2017 Accepted: 30 April 2018 cancer immunotherapy Published: xx xx xxxx Qiyue Hu, Xin Ye, Xiangdong Qu, Dongbing Cui, Lei Zhang, Zhibin Xu, Hong Wan, Lianshan Zhang & Weikang Tao Interleukin-15 (IL-15) can promote both innate and adaptive immune reactions by stimulating CD8+/ CD4+ T cells and natural killer cells (NK) while showing no efect in activating T-regulatory (Treg) cells or inducing activation-associated death among efector T cells and NK cells. Thus, IL-15 is considered as one of the most promising molecules for antitumor immune therapy. To improve the drug-like properties of natural IL-15, we create an IL-15-based molecule, named P22339, with the following characteristics: 1) building a complex of IL-15 and the Sushi domain of IL-15 receptor α chain to enhance the agonist activity of IL-15 via transpresentation; 2) through a rational structure-based design, creating a disulfde bond linking the IL-15/Sushi domain complex with an IgG1 Fc to augment its half-life. P22339 demonstrates excellent developability, pharmacokinetic and pharmacodynamic properties as well as antitumor efcacy in both in vitro assessments and in vivo studies. It signifcantly suppresses tumor growth and metastasis in rodent models, and activates T efector cells and NK cells in cynomolgus monkey. Overall, these data suggest that P22339 has a great potential for cancer immunotherapy. Interleukin 15 (IL-15) is a cytokine of about 12–14 KD discovered by Grabstein et al. in 19941. It is a growth factor for T cells and NK cells, and plays an important role in the development, proliferation and activation of these immune cells. IL-15 has been identifed by the American National Cancer Institute as one of the most prom- ising immunotherapy targets for cancer2. A related cytokine IL-2 has been approved by the US Food and Drug Administration for the treatment of metastatic renal cell carcinoma and malignant melanoma. However, its appli- cation is limited due to its stimulatory efects on Treg cells and its induction of the Activation-Induced Cell Death (AICD) among T efect cells and NK cells. Both processes can limit the memory T cell response and induce T cell tolerance. Both IL-15 and IL-2 bind to the IL-15 receptor (IL-15R) β and γc. Unlike IL-2, IL-15 neither interacts with Treg cells nor induces AICD3,4, therefore avoiding the liabilities of IL-2. IL-15 binds to the IL-15R which consists of α, β, and γc chains. IL-15Rβ (also known as IL-2Rβ or CD122) and IL-15Rγc (also known as CD132) can bind to both IL-15 and IL-2 with intermediate afnity. IL-15Rα is widely expressed and only binds IL-15 with high afnity. IL-15Rα contains a Sushi domain (1–65 amino acids), which is responsible for interacting with IL-15, and is essential for mediating the biological function of IL-155. Although IL-15 has a great potential for therapeutic use6, natural IL-15 has obvious druggability issues, namely a low biological potency, a short half-life7, and an inferior developability8,9. Terefore, to develop an IL-15 based therapeutic agent, it is essential to enhance its potency, extend its half-life and improve its developability. A number of studies indicate that the complex formed by IL-15 and the Sushi domain of soluble IL-15Rα is signif- icantly stronger than IL-15 alone in stimulating the proliferation of memory CD8+ T lymphocytes and NK cells, as well as maintaining the viability of memory CD8+ T cell through the trans-presentation mechanism7,10–13. Tat suggests creating an IL-15/Sushi domain fusion protein can signifcantly increase potency. It has been reported that a covalent linkage formed between the two binding proteins by disulfde bond can make signifcant contribu- tion to the stability of the complex primarily due to the decrease in the entropy of the unfolded protein14. Rational design of disulfde bond has been widely used to improve the protein stability, facilitate the research of protein Shanghai Hengrui Pharmaceutical Co. Ltd., 279 Wenjing Road, Shanghai, 200245, China. Correspondence and requests for materials should be addressed to Q.H. (email: [email protected]) SCIENTIFIC REPORTS | (2018) 8:7675 | DOI:10.1038/s41598-018-25987-4 1 www.nature.com/scientificreports/ Figure 1. Schematic Drawings of selected IL-15 and Receptor α complex molecules. IL-15 is displayed as orange ball (mutations are indicated in the picture). IL-15 receptor α motif is represented by green half circle (Te number of the amino acids is shown in parentheses). Fc domain is illustrated with gray oval. Intermolecular disulfde bond is represented with solid red line. function, and generate novel therapeutic proteins15–19. In this study, we employed a two-step computational mod- eling approach to engineer a new disulfde bond between IL-15 and the Sushi domain of IL-15Rα based on the published co-crystal structure (20; PDB ID: 2Z3Q). In step 1, we selected the potential mutation sites to pair disulfde bonds. In step 2, we evaluated the energetic efects on the protein stability for the selected residues found in step 1. A subset of the mutation combinations were selected and compared by protein expression, in vitro functional assay and stability. L52C of IL-15 and S40C of IL-15Rα were chosen as the fnal mutation combination based on the modeling result. Tere are multiple ways to improve the half-life of a protein molecule21. We chose to fuse the Sushi domain with the Fragment Crystallizable region (Fc) of a human IgG1, and generated a Fc-fusion molecule for improving the PK properties. Te resulted molecule, P22339, was then evaluated with the biological activities, developability, pharmacokinetic and pharmacodynamic properties. Te in vivo studies demonstrated the antitumor immunity of P22339 through the induced proliferations of NK cells, CD8+ and CD4+ T cells which resulted in tumor inhibition. Results Design of IL-15 analogs for improving biological activities and developability. Tere have been many companies or institutions engaging in researches related to IL-15 immunotherapy22–27. Some of the repre- sentative IL-15 and Receptor α complex molecules are shown in Fig. 1. In order to mitigate those liabilities of IL-15 mentioned above, we designed our molecules from three perspectives: (1) Form complex with IL-15Rα to take advantage of the activity boosting efect; (2) Increase molecular weight to pass 50 KD with Fc fusion to reduce the renal fltration and degradation21; (3) Introduce intermolecular disulfde bond between IL-15 and its receptor α to further increase the molecular stability and bioactivity as well as improve the developability of the molecule. Design and Selection of the disulfde bond pairs. We started from the crystal complex structure of human IL-15 and the receptor α, β, γ (28; PDB ID: 4GS7). It could be seen that the three receptors bind to IL-15 in three dif- ferent orientations respectively (see Supplementary Fig. S1). Using 6Å as the cutof, the residues located on the interface of IL-15 and receptor α were identifed (see Supplementary Table S1). We hypothesized that we could modify the residues presented on the interface between IL-15 and receptor α while maintaining the binding of IL-15 to receptors β and γ. Computational Disulfde Scan was performed at the interface between IL-15 and receptor α. Total 9 pairs of Cysteine mutations were selected (Fig. 2a). In order to further evaluate the stability of the mutations, in silico ala- nine scan was carried on all of the residues from the previous Disulfde Scan step. Te calculated mutation energy changes (Fig. 2b) showed that L52A on IL-15 and S40A on IL-15Rα minimally afect the structural stability. Terefore, L52 from IL-15 and S40 from the IL-15Rα were considered as the preferred candidates for mutation engineering to Cysteine as shown in Fig. 2c. Based on the computational results, eight mutational pairs were selected for cell expression. Te specifc co-expression combinations were shown in Supplementary Table S2. Tere were two groups with Cysteine muta- tion indicated in the name, 1) combinations 1–9, IL-15-6His co-expressed with IL-15Rα-Fc fusion molecule, and 2) combinations 10–18, IL-15-Fc fusion molecule co-expressed with IL-15Rα. Te cell supernatant obtained from the co-expression was subjected to western analysis. Te pairing of Fc-fused IL-15Rα and IL-15-6His could result in the correct productions of the intended complex molecule. Te combination 5, 6, 7 showed the best result in terms of the expression levels, with single product in the desired MW range (see Supplementary Fig. S2a). At the same time, the co-expressions of Fc-fused IL-15 and IL-15Rα were prone to result in mismatch, and reduced the amount of the correctly paired target products (see Supplementary Fig. S2b). Te results were in good agreement with the computational prediction. Based on the results above and the yields in CHO cell system, combination #6, L52C-S40C, was named as P22339 and selected for further characterizations. SCIENTIFIC REPORTS | (2018) 8:7675 | DOI:10.1038/s41598-018-25987-4 2 www.nature.com/scientificreports/ Figure 2. Structure-based inter-molecular disulfde bond design. (a) Selected disulfde bond pairs from Disulfde Scan. (b) Result from the In silico Alanine Scanning. Te residues for the preferred amino acid pair are indicated in blue. (c) IL-15 is colored in orange and the receptor α is colored in green. Te intra-molecular disulfde bonds within each molecule are rendered in stick mode.